Boosts quality and cuts costs

Today’s advanced hybrid electric vehicles (HEVs) bring together an almost unimaginable collection of mechanical components, electronic control systems, wiring harnesses and interfaces. For this complex machine to function properly, all components, coupled with other elements for safety, comfort and efficiency, must work together seamlessly as a reliable, cost-effective drivetrain. The challenge lies in building, analyzing and debugging these separate components so that when they are assembled and finally shipped, a reliable vehicle is ready for the road.

By David Smith, Synopsys Scientist, Saber Product Line, Hillsboro, Oregon

Traditionally, the components and subsystems in a drivetrain were developed and tested almost in a vacuum; each team of mechanical, electrical and software engineers worked separately designing, building and testing subsystem components against a set of allowable defect tolerances. The problem with this method was that any given component in the drivetrain might work within tolerable limits in isolation, but could fail when integrated with other drivetrain components. Even if each subsystem met its assigned tolerance requirements, the combination of individual allowable defects could lead to failure when coupled with other components operating within their tolerance ranges. This combined failure resulted in warranty repairs, rounds of finger pointing by suppliers and OEMs and bad publicity for the manufacturer.

Virtually, the right approach

For a complex design, like an HEV, a Robust Design methodology consisting of virtual prototyping and simulation is essential. With this methodology, a system’s design, analysis and debugging can be controlled by engineers at workstations, as opposed to traditional methods where physical prototypes are used. Prototypes are limited by cost because a manufacturer must develop a relatively small number to stay within tight budget guidelines. Additionally, prototyping in extreme environmental conditions can be impractical or dangerous.

In a Robust Design environment, however, tens of thousands of permutations of values can be applied to drivetrain components at minimal incremental cost (and no danger!); this permits engineers to analyze a wide range of variations (due to environment or part tolerance) for a given component and their overall effect when combined with other elements in a design. This process greatly reduces the chance of failures in the field, where all of the elements must work together in harsh real-world conditions.

There are numerous simulation and statistical analysis tools on the market that employ proven models, design languages, component libraries and standards. The problem lies in educating mechanical, electrical and software engineers on proper Robust Design methodologies and best practices for debugging virtual prototypes. Currently, teams in each discipline are familiar with the prototyping and testing methodologies in their separate domains, and are well-versed in using tools that are meant for each particular subsystem. Most mechanical simulation, for example, uses finite element analysis tools, not dynamic simulators. Similarly, software engineers have their own tools and methodologies to check their code for inefficiencies and bugs that can cause system failures. For their part, electro-mechanical engineers rely mainly on Spice simulators for the analog components in their designs.

While these tools and methodologies have worked well in the past, they are inadequate for spanning the disparate realms of electric, mechanical and hydraulic systems that converge in an HEV drivetrain (Figure 1). An overarching multi-domain simulation, analysis and debug methodology is therefore essential for a successful HEV design.
 

Figure 1: In a virtual prototyping and simulation environment, each component in the drivetrain can be created, exercised and analyzed before being integrated and simulated as part of the whole design

Complex challenges demand proven solutions
Software tools, such as Synopsys’ Saber circuit simulator for mixed-signal, mixed-domain Mechatronic systems, have been used successfully for system verification for some time now; they are perfectly suited to the growing market requirement for Zero Defect design of complex new target applications like the HEV drivetrain.   Saber’s analysis and modeling capabilities, comprehensive model libraries and multi-language model creation tools help engineers achieve an optimized, robust design. Saber also includes failure mode effects analysis (FMEA), which can analyze virtual prototypes of any system under various conditions from ideal to worst case.  These advanced simulation environments give engineers the ability to simulate, analyze and verify interactions between multiple physical domains, allowing them to factor electrical, magnetic, mechanical, thermal and hydraulic effects into their design verification processes.

Simulating a virtual prototype of a system begins with the construction of a dynamic model that will accurately represent the behavior of a system across the operating conditions it will face. Hardware description languages (HDLs), like VHDL-AMS and MAST, allow engineers to create complex models that span engineering disciplines. These models go well beyond capturing nominal system behavior and include parameter tolerances to account for the variance and uncertainty that occur in the physical world (Figure 2).  
 

Figure 2: In this VHDL-AMS model of a lithium-ion battery, statistical tolerances can be analyzed and represented in a waveform, allowing engineers to test the design under a range of operating conditions

Properly developed virtual models can also be employed in any combination of statistical, parametric, sensitivity, stress and failure mode analyses to promote complete system verification. Statistical analysis can predict how component tolerance variations affect system performance, allowing designs to achieve Six Sigma quality. Parametric analysis lets engineers fine-tune key parameters in a design, while sensitivity analysis can determine which parameters most affect system performance. Stress analysis helps engineers evaluate the degree of component stress in a system during operation. The aforementioned failure mode analysis plays a key role as well, by providing a vehicle for analysis and assessment of systems under various user-defined fault conditions. Post-simulation, automatically generated reports help designers quickly assess the reliability of the complete system design (Figure 3).
 

Figure 3: A virtual prototyping environment lets engineers achieve test coverage that would be unattainable with actual physical models. Superior results can be achieved when stress analysis, worst case scenarios, signal integrity and other conditions are addressed and analyzed

As these individual component models are integrated together, statistical simulation techniques (e.g. Monte Carlo analysis) can be used to predict the impact of part variations in the system and further identify the components and parameters that have the greatest effect on system robustness. Early in the product design, possibly even prior to prototyping, issues due to multi-domain tolerance stackup can be addressed and verified (Figure 4). Similarly, product reliability may be quantified, through either injecting faults into the system, or by identifying worst-case performance of the virtual prototype that might be impractical (or impossible) to measure on a physical prototype.
 

Figure 4: As each component is integrated into the drive train, statistical simulation lets engineers analyze and understand the impact of a range of tolerances within each component and their combined effect on the whole system


Standards bodies offer open solutions

When selecting the right simulation environment, engineers must also take standards efforts into account. Using tools and methodologies based upon open standards can minimize risk for engineers without limiting their choice of vendors or library models. Standards bodies, such as the Society for Automotive Engineers (SAE) International and the German Automobile Industry Association (VDA), are resources that designers may tap for useful information on standardization efforts for design languages, model library interoperability and other related concerns. Additionally, focused working groups within these bodies, such as the VDA’s FAT-AK30, can also be useful for engineers who want to track and possibly participate in standardization efforts. (For more information, visit: http://www.sae.org/servlets/index; http://www.vda.de/en/verband/index.html; and http://fat-ak30.eas.iis.fraunhofer.de/index_en.html).

The time is now

Traditional design, debug and analysis tools are quickly running out of steam, as the stringent requirements for product quality (e.g. Zero Defect Design) are applied to ever more complex applications. Subsystems and components that are built and tested as separate entities are more likely to fail when employed as a single working unit in a complex HEV. Companies that evaluate and use tools and methodologies geared towards Robust Design processes will lower costs, reduce risk and ultimately gain market share with vehicles that move passengers efficiently and safely to their destinations.

www.synopsys.com